FIDE Ethics Commission ruling on ICGA/Rybka complaint

[lucasart]

Why do people keep discussing about whether or not SF will participate?
I thought this was a discussion about Rybka

SF will not participate in this fake world championship:
https://groups.google.com/forum/?fromgr ... F2VYzHEsR4

Maybe things were different in the 80s, but today in 2015, ICGA is useless, irrelevant, and incredibly arrogant. We do not need the ICGA. It's the ICGA needs SF desperately to save their pathetic tournament from falling into oblivion.

[michiguel]

The criteria to be used is "Claims that are proven to the satisfaction of the ICGA"

So you are afraid that it can be proven to the satisfaction of the ICGA that there is stolen code if in fact there is none, and the source code is openly available to anyone? You do realize that this sounds like a rather extreme case of paranoia?

I am not afraid of anything bad could happen. It already did. I was just pointing the fact that "shaming" had a lower standard, which is satisfying an institution whose rules are not even written properly, are a moving target, and their procedures have changed constantly. It was also a fact that ICGA could not carry a decent procedure and violated rules from their umbrella organization (FIDE) and even its own standards. For instance, they promised confidentiality, while a leading member of the secretariat was leaking info to the public.

In this scenario, why would anybody enter a program just to test what happens?

Miguel

[michiguel]

Why do people keep discussing about whether or not SF will participate?
I thought this was a discussion about Rybka

It is a discussion about ICGA and its rules. SF is just a case study of what could happen.

Miguel

SF will not participate in this fake world championship:
https://groups.google.com/forum/?fromgr ... F2VYzHEsR4

Maybe things were different in the 80s, but today in 2015, ICGA is useless, irrelevant, and incredibly arrogant. We do not need the ICGA. It's the ICGA needs SF desperately to save their pathetic tournament from falling into oblivion.

[bob]

I've added some bold text. ICGA is a tournament for programmers. TCEC is simply about the best engine from a list selected by Martin.

They agreed to it by drawing up the license agreement. Nothing to pretend there, it is all written in black and white.

It is utterly insulting and nonsensical to interpret the release of code under the GPL as agreement of the authors to participate as programmers in some tournament.

The GPL does not say "we agree to our participation in ICGA tournaments". In this respect the GPL does not differ one iota from any other license. Why should the GPL be interpreted as "we agree to participating in ICGA tournaments" but, for example, Komodo's license not?

Actually the GPL allows you to do ANYTHING with the source. Main stipulation is that if you release/sell/distribute the resulting executable, you have to also do the same for the source code. There are NO limits placed on how/where/when the code can be used, however.

Sure, but what does the source have to do with this? You don't need the source to compete in a tournament. I can download a binary of Crafty and play against it on my PC.

If I'm allowed to run an executable, I can run it in a test environment such as TCEC or an ICGA tournament. That is, the license allows that.

The point, that seems hard to grasp, is that ICGA is a programmer's tournament where programmer's compete in their personal capacity (possibly represented by someone they appointed, but that makes no difference).

To purport that Marco/Joona/Tord agreed to competing in an ICGA tournament by releasing SF under the GPL, or to purport that Mark and Larry agreed to competing in an ICGA tournament by releasing Komodo under a license that is silent on testing Komodo against other engines, is nonsensical.

With TCEC there is no such problem, as TCEC does not purport that the programmer's agreed to personal participation (even though Martin seems to voluntarily respect the express wishes of authors that prefer their engine to stay out of TCEC).

My only comment was that the GPL doesn't prevent anything regarding the use of the code, and only affects the distribution.

The issue you raise would not be GPL, rather it would be ICGA. Crowd-sourced software development is something fairly new. But when you have such, you generally have one (or two or three) people at the top of the food chain. Linus for example in regard to Linux. For me, the three usual people are the authors, and they should have the final say in whether their code is entered in the ICGA event or not, and the ICGA should listen to the three of them rather than get into the GPL debate or anything else.

I can see lots of issues here. One classic was Murray Campbell. Started out on the HiTech project, then moved to the competing deep thought (CMU) project and then followed with the Deep Blue (IBM) project. Where does he factor in? IMHO he is Deep Blue, his LAST affiliation. Had he continued to work on the HiTech project as well, then things get messier.

For a crowd-sourced program do you need ALL contributors to agree? I think that is too stringent a requirement. One that might not be possible, period. So "the primary author (or authors)" ought to be the one(s) that control this. And I think the ICGA would go in that direction if there were a conflict.

The stickier issue is the derivative problem, i.e. Fruit and Toga, or Crafty and Bionic or whatever, as examples. You really don't want two of the same program, even if one has been modified. Or do you? I don't, but I'm willing to go along with the crowd. But the minute there is more than one, it will attract my attention. And should I look around and see 2-3-4 robbolito derivatives entered, I would certainly choose to not participate myself. Everyone else would be free to do whatever they wanted of course. And if most events allow all the robbos then I would have to make a strategic choice - accept all the clones or not participate any more. Neither is an optimal circumstance.

So the ICGA does need to get ready for this issue. If not with StockFIsh then with the next "newest and strongest" whatever that might be. I do think it makes sense to have a "core group" (one or more primary authors) that are considered responsible for the project, and they are given the right to make the decision to participate or not participate, and the various contributors just go along for the ride or move on.

[syzygy]

Why do people keep discussing about whether or not SF will participate?
I thought this was a discussion about Rybka

Hmmm, it seems this thread has been drifting a bit

SF will not participate in this fake world championship:
https://groups.google.com/forum/?fromgr ... F2VYzHEsR4

Interesting thread, I had missed it. The opening post refers to this thread which I haven't read where apparently essentially the same was said as what I wrote before:

Some members of the Stockfish community (i.e. Marco, Tord, Lucas, ...) expressed that they aren't interested in this event at all, since it is of no statistical relevance. They clearly stated that an official registration of the Stockfish team in ICGA events will never happen. But the bottom line was that there would be no problem whatsoever if somebody else would play with Stockfish in the WCCC, as long as it is beyond question that the Stockfish team has nothing to do with it. At least that was my interpretation of the discussion.

Exactly that phrase in bold is what the Stockfish team can require. For the rest it is up to the ICGA and up to the participants that are suddenly confronted with a participant not bringing his own program, but a big hammer that will likely wipe the floor with the rest.

[bob]

Why do people keep discussing about whether or not SF will participate?
I thought this was a discussion about Rybka

SF will not participate in this fake world championship:
https://groups.google.com/forum/?fromgr ... F2VYzHEsR4

Maybe things were different in the 80s, but today in 2015, ICGA is useless, irrelevant, and incredibly arrogant. We do not need the ICGA. It's the ICGA needs SF desperately to save their pathetic tournament from falling into oblivion.

I think that in this case, the ONLY opinion that will hold any weight will be the opinion(s) of the three primary authors ONLY.

[bob]

I can enter a modified version of SF, use SF code since it is legal to use, name developers I see in the SF team and that should be the only requirement. It does not say permission is needed from them.

Indeed, the rules seem to allow this. But I am pretty sure this is by mistake. Of course the written permission would often be required to make the use legal. (Not so in the case of the GPL, though.) The clause was probably added with things like tablebase probing in mind. I don't think they realized that it would punch a whole large enough to sink the ship...

In fact, I cannot see how this rule forbids to use a program that searches and requests the eval from a another binary, as long as the authors are listed.

This is exactly what GridChess was. It used Toga II binaries running on many different machines, and the original part of the program just coordinated their searches.

I cannot even see (but I may have missed it) that an author cannot be listed in two different teams. Actually, it must be allowed since I read this

"Registration form
We changed the contribution fee for the Olympiad and Chess events:
Participation: 50 Euro for the first program, 25 Euro for the second, third, etc. program.

Use one form for each program. Multiple programs are allowed for one person."

This is for the Olympiad, not the WCCC/WCSC. Multiple programs here means programs for different games. Like I am participating in the Xiangqi, Shogi, mini-Shogi and Tori Shogi.

Yes, but that is just an interpretation.

Miguel

I believe the general feeling here is that there will be exactly ONE entrant from a "family". There were such issues in the past and this has generally been the accepted policy. If someone wants to enter a program derived from Fruit, then they need Fabien's permission, and if given, Fruit will be excluded from the event to allow the derivative (one member from the family) to participate.

[syzygy]

My only comment was that the GPL doesn't prevent anything regarding the use of the code, and only affects the distribution.

The issue you raise would not be GPL, rather it would be ICGA.

Indeed, and I think I've been quite clear about that myself.

Crowd-sourced software development is something fairly new. But when you have such, you generally have one (or two or three) people at the top of the food chain. Linus for example in regard to Linux. For me, the three usual people are the authors, and they should have the final say in whether their code is entered in the ICGA event or not, and the ICGA should listen to the three of them rather than get into the GPL debate or anything else.

I fully agree that, given the usual setup of ICGA tournaments, this is by far the most reasonable position the ICGA could take regarding participation of Stockfish. Marco, Joona and Tord are universally accepted as "the authors" of Stockfish.

For a crowd-sourced program do you need ALL contributors to agree? I think that is too stringent a requirement. One that might not be possible, period. So "the primary author (or authors)" ought to be the one(s) that control this.

Agreed.

And I think the ICGA would go in that direction if there were a conflict.

There seem to be indications that the ICGA was willing to accept an non-SF developer (and for sure a non-primary developer) to participate using SF.

Personally I am entirely in agreement with you that it would go against the spirit of ICGA tournaments to allow this. On the other hand, the ICGA does have the right to allow it (provided the other participants do not object), but only if it is made clear that the participation of SF is not authorised by the 3 primary developers.

The stickier issue is the derivative problem, i.e. Fruit and Toga, or Crafty and Bionic or whatever, as examples. You really don't want two of the same program, even if one has been modified. Or do you?

I have never competed in iCGA tournaments so I do not consider that I am entitled to want anything, but I feel a lot of sympathy for not allowing two nearly identical engines to enter. But clearly there is no engine nearly identical to SF that would have a better right of entering than SF. (However I am not trying to apply Rule 2 here. For example Glaurung is not "nearly identical" to SF in my book, so I would have no objection to it entering. Whether Tord would be allowed to participate twice is a different issue on which I have no strong opinion.)

[Roger Brown]

I fully agree that, given the usual setup of ICGA tournaments, this is by far the most reasonable position the ICGA could take regarding participation of Stockfish. Marco, Joona and Tord are universally accepted as "the authors" of Stockfish.

Hello Ronald,

I would accept that these three would have the "right" (let us take a wider interpretation than the strict, legal or procedural definition of that word) as authors of Stockfish to enter - or refuse to enter - rather than some other person or group.

Given the above, I hope you can appreciate that this question is not intended to be mischevious at all. Do you not consider your own additions to Stockfish as significant? Should the answer be no, why not? Just curious.

Later.

[syzygy]

I would accept that these three would have the "right" (let us take a wider interpretation than the strict, legal or procedural definition of that word) as authors of Stockfish to enter - or refuse to enter - rather than some other person or group.

Well worded.

Given the above, I hope you can appreciate that this question is not intended to be mischevious at all. Do you not consider your own additions to Stockfish as significant? Should the answer be no, why not? Just curious.

Thanks for asking.

When I released the TB code, my "SF addition" was intended only to show other engine programmers how to integrate the TB access code into their engine. I figured that most engine developers would be at least somewhat familiar with the Stockfish code, so using SF as example seemed to make sense. It was clear from the start that Marco would never accept it (and I never offered it), so I did not consider it a contribution to SF.

Eventually SF leadership changed a bit, so after Houdini and Komodo, now also the official SF releases contain the code. In my view I simply released the code to anyone interested and the SF developers rightfully took it and integrated it, just like the Houdini and Komodo developers did (among others).

I have to admit that I got somewhat annoyed some months ago when some SF developers were discussing a "simplification" of the TB code that would break its logic, but perhaps in a way that would not show in the usual ultrashort tc tests (so then you get Marco arguing that he doesn't care whether theory predicts that 1+1=2 if "hard evidence" (= statistical test results) shows otherwise, but see this thread that I was just reading... so sad that it is funny). At the same time I told them that it is entirely up to them what they decide to do with it. If you ask me, it is a question of time until Marco gets the upper hand again and the TB code is ripped out.

So no, I don't consider that I have significantly added to SF. (And certainly in Elo terms it is insignificant.)

Btw, even if I had developed the TB code specifically for SF, my contribution would not give me a title to calling myself a "primary developer". The number of code lines might be significant, but it hardly touches the real engine which is what makes Stockfish Stockfish.